The AVG provides a simpler means
of automatically convolving with a triangular line shape, which is the
recommended option if all you need
is an output grid which represents the average values over spectral intervals
rather than discrete samples of the spectrum.
The incompatibility with the
OPT flag
is ideological: instruments do not measure optical thickness directly,
so spectral convolution or averaging
in 'optical thickness' is meaningless. If you really
do want something representing convolved optical thickness
(and are you sure you do?) I suggest using the RFM to write the
unconvolved optical
thickness spectrum and convolving it yourself.